Signal to background discrimination for the production of double Higgs boson events via vector boson fusion mechanism in the decay channel with four charged leptons and two b-jets in the final state at the LHC experiment

TL;DR

This paper explores the use of deep learning techniques to improve the discrimination between signal and background events in double Higgs boson production via vector boson fusion at the LHC, aiming to enhance detection sensitivity.

Contribution

It introduces a deep neural network approach with hyper-parameter optimization for better signal-background separation in a rare Higgs production process.

Findings

Deep neural network achieves high discrimination performance.

Hyper-parameter scanning improves training efficiency.

Method enhances detection prospects of double Higgs events.

Abstract

At the CERN Large Hadron Collider experiment, the non-resonant double Higgs production via vector-boson fusion represents a unique mean to probe the VVHH (V=Z, W$^{\pm}$) Higgs self-coupling at the current center of mass energies. Such a rare signal cannot be separated efficiently from huge backgrounds by applying a few-observables cut-based selection. Indeed, in this work, a Deep Learning algorithm is used to decide whether an event is more signal- or background-like. In particular, we report on two main aspects: results of a hyper-parameters parallel scanning strategy to distribute the training process across multiple nodes on the ReCaS-Bari data center computing resources and the discriminating performance of a Deep Neural Network architecture.